Abstract Introduction Early identification of whether bloodstream infections (BSIs) are caused by gram-negative or gram-positive bacteria is crucial for initiating timely and appropriate antibiotic treatment. This study aimed to investigate potential laboratory markers for predicting the type of bacterial infection in patients with suspected BSI. Methods Patients who presented at the emergency department with positive blood cultures were included in this study. Clinical and laboratory data collected within the first 24 hours of admission were analyzed. Results The study included 338 patients. Clinically significant differences in platelet count and procalcitonin (PCT) level were observed between patients with gram-positive and gram-negative infections. Platelet counts and PCT levels as predictors of infection were relatively ineffective, however, with receiver operating characteristic (ROC) area under the curve (AUC) values of 0.63 and 0.66, respectively. Subsequently, a specific cluster of patients characterized by normal to high platelet counts (170-318 ×109/L) and mildly elevated PCT levels (0-31.06 ng/mL) was identified using k means clustering. In this cluster, platelet counts of 284 ×109/L or higher combined with PCT levels of 2.73 ng/mL or less demonstrated a sensitivity of 70.3%, a specificity of 78.9%, and ROC AUC values of 0.82 for predicting gram-positive bacterial infections. Discussion The combined use of platelet count and PCT level could predict gram-positive bacterial infections in specific populations. This study may provide a practical strategy for the rapid identification of bacterial groups in patients with suspected BSI.
Wang et al. (2025) studied this question.